This paper examines whether and to what extent changes in beliefs about own HIV status induce changes in risky sexual behavior using data from married males living in three regions of Malawi. Risky behavior is measured as the propensity to engage in extramarital affairs. The empirical analysis is based on panel surveys for years 2006 and 2008 from the Malawi Diffusion and Ideational Change Project (MDICP), which contain detailed information on beliefs about HIV status and on sexual behaviors. Many individuals change their beliefs over time, because of opportunities to get tested for HIV and informational campaigns. We estimate the effect of belief revisions on the propensity to engage in extra-marital affairs using a panel data estimator developed by Arellano and Carrasco (2003), which accommodates unobserved heterogeneity as well as belief endogeneity arising from the dependence of current beliefs on lagged behaviors. We find that downward revisions in the belief of being HIV positive lead to an increased propensity to engage in extra-marital affairs and upward revisions to a decreased propensity. The estimates are shown to be robust to underreporting of affairs. Using our estimates and a standard epidemeological model of disease transmission, we find that increasing the responsiveness of beliefs to test results would reduce the HIV transmission rate as infected individuals reduce sexual behavior and decrease the likelihood that uninfected persons have contact with an HV-positive person.